Add '0xCAFE' extra field to the first entry of the output jar (which is META-INF/). Unix 'file' utility uses it to distinguish jar file from zip file.

--
MOS_MIGRATED_REVID=134102453

4 files changed, 35 insertions, 10 deletions

diff --git a/src/tools/singlejar/output_jar.cc b/src/tools/singlejar/output_jar.cc
index 855afe6d2c..d5e2c10969 100644
--- a/src/tools/singlejar/output_jar.cc
+++ b/src/tools/singlejar/output_jar.cc
@@ -191,7 +191,7 @@ int OutputJar::Doit(Options *options) {
   bool compress = options_->force_compression || options_->preserve_compression;
   // First, write a directory entry for the META-INF, followed by the manifest
   // file, followed by the build properties file.
-  AddDirectory("META-INF/");
+  WriteMetaInf();
   manifest_.Append("\r\n");
   WriteEntry(manifest_.OutputEntry(compress));
   if (!options_->exclude_build_data) {
@@ -538,8 +538,8 @@ void OutputJar::WriteEntry(void *buffer) {
     diag_err(1, "%s:%d: write", __FILE__, __LINE__);
   }
   // Data written, allocate CDH space and populate CDH.
-  CDH *cdh = reinterpret_cast<CDH *>(
-      ReserveCdh(sizeof(CDH) + entry->file_name_length()));
+  CDH *cdh = reinterpret_cast<CDH *>(ReserveCdh(
+      sizeof(CDH) + entry->file_name_length() + entry->extra_fields_length()));
   cdh->signature();
   // Note: do not set the version to Unix 3.0 spec, otherwise
   // unzip will think that 'external_attributes' field contains access mode
@@ -555,7 +555,7 @@ void OutputJar::WriteEntry(void *buffer) {
   TODO(entry->uncompressed_file_size32() != 0xFFFFFFFF, "Handle Zip64");
   cdh->uncompressed_file_size32(entry->uncompressed_file_size32());
   cdh->file_name(entry->file_name(), entry->file_name_length());
-  cdh->extra_fields(nullptr, 0);
+  cdh->extra_fields(entry->extra_fields(), entry->extra_fields_length());
   cdh->comment_length(0);
   cdh->start_disk_nr(0);
   cdh->internal_attributes(0);
@@ -565,9 +565,18 @@ void OutputJar::WriteEntry(void *buffer) {
   free(reinterpret_cast<void *>(entry));
 }
 
-void OutputJar::AddDirectory(const char *path) {
+void OutputJar::WriteMetaInf() {
+  const char path[] = "META-INF/";
   size_t n_path = strlen(path);
-  size_t lh_size = sizeof(LH) + n_path;
+
+  // META_INF/ is always the first entry, and as such it should have an extra
+  // field with the tag 0xCAFE and zero bytes of data. This is not the part of
+  // the jar file spec, but Unix 'file' utility relies on it to distiguish jar
+  // file from zip file. See https://bugs.openjdk.java.net/browse/JDK-6808540
+  const uint8_t extra_fields[] = {0xFE, 0xCA, 0, 0};
+  const uint16_t n_extra_fields =
+      sizeof(extra_fields) / sizeof(extra_fields[0]);
+  size_t lh_size = sizeof(LH) + n_path + n_extra_fields;
   LH *lh = reinterpret_cast<LH *>(malloc(lh_size));
   lh->signature();
   lh->version(20);  // 2.0
@@ -577,7 +586,7 @@ void OutputJar::AddDirectory(const char *path) {
   lh->compressed_file_size32(0);
   lh->uncompressed_file_size32(0);
   lh->file_name(path, n_path);
-  lh->extra_fields(nullptr, 0);
+  lh->extra_fields(extra_fields, n_extra_fields);
   known_members_.emplace(path, EntryInfo{&null_combiner_});
   WriteEntry(lh);
 }
diff --git a/src/tools/singlejar/output_jar.h b/src/tools/singlejar/output_jar.h
index 082bf7a33b..be3bef8aad 100644
--- a/src/tools/singlejar/output_jar.h
+++ b/src/tools/singlejar/output_jar.h
@@ -71,8 +71,8 @@ class OutputJar {
   off_t Position();
   // Write Jar entry.
   void WriteEntry(void *local_header_and_payload);
-  // Write a directory entry.
-  void AddDirectory(const char *path);
+  // Write META_INF/ entry (the first entry on output).
+  void WriteMetaInf();
   // Append given Central Directory Header to CEN (Central Directory) buffer.
   CDH *AppendToDirectoryBuffer(const CDH *cdh);
   // Reserve space in CEN buffer.
diff --git a/src/tools/singlejar/output_jar_simple_test.cc b/src/tools/singlejar/output_jar_simple_test.cc
index de9751ba15..6bad8e3fc4 100644
--- a/src/tools/singlejar/output_jar_simple_test.cc
+++ b/src/tools/singlejar/output_jar_simple_test.cc
@@ -115,9 +115,12 @@ TEST_F(OutputJarSimpleTest, Empty) {
   CreateOutput(out_path, {});
   InputJar input_jar;
   ASSERT_TRUE(input_jar.Open(out_path));
+  int entry_count = 0;
   const LH *lh;
   const CDH *cdh;
+  const uint8_t cafe_extra_field[] = {0xFE, 0xCA, 0, 0};
   while ((cdh = input_jar.NextEntry(&lh))) {
+    ++entry_count;
     ASSERT_TRUE(cdh->is()) << "No expected tag in the Central Directory Entry.";
     ASSERT_NE(nullptr, lh) << "No local header.";
     ASSERT_TRUE(lh->is()) << "No expected tag in the Local Header.";
@@ -163,6 +166,19 @@ TEST_F(OutputJarSimpleTest, Empty) {
     EXPECT_GE(now, entry_time) << now_time_str << " vs. " << entry_time_str;
     EXPECT_LE(now, entry_time + 300) << now_time_str << " vs. "
                                      << entry_time_str;
+
+    // The first entry should be for the META-INF/ directory, and it should
+    // contain a single extra field 0xCAFE. Although
+    // https://bugs.openjdk.java.net/browse/JDK-6808540 claims that this extra
+    // field is optional, 'file' utility in Linux relies on to distinguish
+    // jar from zip.
+    if (entry_count == 1) {
+      ASSERT_EQ("META-INF/", lh->file_name_string());
+      ASSERT_EQ(4, lh->extra_fields_length());
+      ASSERT_EQ(0, memcmp(cafe_extra_field, lh->extra_fields(), 4));
+      ASSERT_EQ(4, cdh->extra_fields_length());
+      ASSERT_EQ(0, memcmp(cafe_extra_field, cdh->extra_fields(), 4));
+    }
   }
   input_jar.Close();
   string manifest = GetEntryContents(out_path, "META-INF/MANIFEST.MF");
diff --git a/src/tools/singlejar/zip_headers.h b/src/tools/singlejar/zip_headers.h
index 02f8167973..226c7cb800 100644
--- a/src/tools/singlejar/zip_headers.h
+++ b/src/tools/singlejar/zip_headers.h
@@ -243,7 +243,7 @@ class LH {
   uint8_t *extra_fields() {
     return reinterpret_cast<uint8_t *>(file_name_) + file_name_length();
   }
-  void extra_fields(uint8_t *data, uint16_t data_length) {
+  void extra_fields(const uint8_t *data, uint16_t data_length) {
     extra_fields_length_ = htole16(data_length);
     if (data_length) {
       memcpy(extra_fields(), data, data_length);